Shiming Song

Papers

Total Citations

H-Index

About

Shiming Song is a rising researcher in computer vision and 3D scene understanding, with a focus on panoramic perception and reconstruction. His most notable contribution, **Pano3R**, introduces a training-free framework for panoramic 3D reconstruction, directly addressing the critical challenge of adapting pinhole-based methods to 360° inputs. By eliminating the need for retraining on scarce panoramic datasets, this work offers a scalable solution for immersive applications in robotics, augmented reality, and autonomous driving. Though recently published in 2025, Pano3R has already garnered early citations, signaling its potential to reshape how 360° visual data is processed. Song’s research bridges a key gap between classical 3D reconstruction and emerging panoramic sensors, enabling robust scene understanding without the prohibitive cost of large-scale panoramic annotation. His work is particularly valuable for students and engineers seeking efficient, generalizable approaches to real-world spatial AI. As the demand for immersive and autonomous systems grows, Song’s contributions stand out for their practical impact and elegant avoidance of data-hungry training paradigms.